Open xyn0813 opened 2 years ago
We did not observed obvious bias for genes in our data, the gene expression of scTE corelates tighly with cell ranger and STAR in our data.
It maybe data dependent. It is possible as you explain, we use seurat/scanpy/scran for normalization without any sepcial settings. Or may only influence for some specific genes (overlaped to TEs?), can you check how many genes are biased, and how many of them are overlaped to TEs?
I think you may misunderstand my question. Let me be clear :how do you get differentially expressed genes and TEs?
We have tried both, and then plot each of them (will not too many) to see if there are any false discoveires.
when I tried to find differential expressed genes and TEs for different cell clusters using expression matrix containing both genes and TEs, I noticed that I got less differential expressed genes comparing with using expression matrix containing genes only (there is no difference with TEs). I assumed that the counts of TEs are much higher than genes, so some differential expressed genes whose counts is low cannot be found after normalization. Is my assumption resonable?How to normalize data properly? How to solve such a problem?